PHD IN FRESHWATER DIATOM-BASED BIOMONITORING
LEVERAGING MACHINE LEARNING
Keywords
Freshwater benthic diatoms, aquatic ecology, biomonitoring, data analysis, imaging, machine
learning
General context:
Diatoms are ubiquist microalgae inhabiting all aquatic environments. They are widely
used organisms in freshwater ecological assessment (biomonitoring). However, the most
efficient diatom-based diversity indices require a high level of taxonomic knowledge, which
involves time and expert training. Traditional taxonomic identification of these organisms is
based on the observation of both shape and ornamentation of their frustule, a highly ornamented
silica shell, using conventional optical methods (Fig. 1). These morphological features/traits
are not always easy to characterize, depending on the experience of the operator, the quality of
the image and the continuous evolution of diatom classification. Thus, the estimation of
diatom-based diversity indices is often prone to multiple biases, implying that the development
of a genuine tool is needed for supporting in decision-making in diatom identification.
Fig. 1: typical images of different diatom species observed under the microscope (x1000) using
permanent slides.
Automatic classification of diatoms has remained a challenge since the first attempts during
the 90’s, but the recent development of deep learning approaches is promising for solving
image-based diatom classification problems. Our team has recently developed several deep-
learning based models to automatically classify freshwater benthic diatoms at the species level
and quantify key morphological traits. These advancements aim to improve ecological
diagnostics compared to traditional methods by reducing time, cost and variability, therefore
supporting robust management responses to anthropogenic pressures on aquatic ecosystems.
These tools are also promising for addressing more general ecological questions.
Within the recently funded project BIOINDIC-IA ([Link]
we are seeking a motivated PhD candidate to join an interdisciplinary research project
focused on leveraging deep learning and advanced image processing techniques to improve
the current tools for biomonitoring of aquatic ecosystems.
PhD project
The PhD candidate will be in charge of the following tasks:
- Diatom datasets assembly and curation: update of traits dataset (based on bibliography) and
annotated image datasets (based on state-of-the-art image acquisition set-up)
- Data analysis: Explore diatom taxonomic and functional diversity at various spatial and
temporal scales, compare automated approaches with traditional methods based on relevant
case studies which will be representative of various taxonomic and morphologic trait
diversity levels
- Real-world impact: lay the groundwork for implementing these innovative tools within
institutional biomonitoring frameworks for practical applications by contributing to
dedicated activities involving end-users, stakeholders and managers (OFB and Water
Agencies in France, Administration de la gestion de l'eau (AGE) in Luxembourg).
Candidate profile
- Education: You hold a MSc or equivalent in Biology, Environmental Science, or a related
discipline.
- Experience and skills: Applicants should have a solid background in ecology and data
analysis, ideally but not exclusively in aquatic ecology and biomonitoring. Prior experience
in diatom taxonomy; ecology and microscope image acquisition would be a plus. The thesis
project will be carried out in close collaboration with another thesis (already filled) whose
objective will be to develop/improve deep learning models. Therefore, applicants should
like working in a team, explaining and presenting research ideas and results in a strong
interdisciplinary context (ecology, computer vision, machine learning experts) but also a
non-scientific audience (end-users).
- Language skills: Good level of both written and spoken English; knowledge of spoken
French is a plus as to interact with the French-speaking community of end-users.
General information
Financial support is available for 4 years from FNR (Luxembourg National Research Fund).
The position will be located at LIST (Luxembourg Institute of Science and Technology,
Belvaux, Luxembourg); with secondment at LIEC (Laboratoire Interdisciplinaire des
Environnements Continentaux (LIEC, UMR 7360 CNRS-Université de Lorraine, Metz,
France).
The PhD candidate will be co-supervised by Dr Carlos Wetzel (LIST, diatom ecology and
taxonomy, imaging) and Dr Martin Laviale (LIEC, diatom ecology, biomonitoring, AI for
ecology) in close collaboration with computer vision and machine learning experts Dr Cédric
Pradalier (GeorgiaTech Europe, Metz) and Dr Jérémy FIX (LORIA, Metz)
Application
Candidates with the desired profiles should send a CV and a letter of motivation (including the
contact details of previous advisors) to the two supervisors ([Link]@[Link],
[Link]@[Link]). Relevant applications will be selected for further oral interviews. The
position should be filled as soon as possible, applications will be considered until the position
is filled.
Relevant bibliography
Soininen, Jet al. 2016. Global patterns and drivers of species and trait composition in diatoms. Global
Ecology and Biogeography [Link] .
Larras, F. et al. 2017. Assessing anthropogenic pressures on streams: A random forest approach based
on benthic diatom communities. Science of The Total Environment
[Link]
Karimi, B. et al. 2017? Microbial diversity and ecological networks as indicators of environmental
quality. Environ Chem Lett [Link]
Martini, S. et al. 2021 Functional trait-based approaches as a common framework for aquatic
ecologists. Limnology and Oceanography [Link]
Orenstein et al 2022. Machine learning techniques to characterize functional traits of plankton from
image data. Limnology and Oceanography [Link]
Venkataramanan et al 2024 “UDE DIATOMS in the Wild 2024”: a new image dataset of freshwater
diatoms for training deep learning model. Gigascience [Link]
